Identification of colloidal haze protein in Chinese rice wine (Shaoxing Huangjiu) mainly by matrix‐assisted laser ionization time‐of‐flight mass spectrometry

Abstract As one of the three most famous brewed wines in the world, Chinese rice wine is made from rice and husked millet, containing 14 percent to 20 percent alcohol. Highly original, yellow wine brewing techniques are regarded as the model of the wine brewing industry in Asia. Shaoxing Huangjiu is produced in Zhejiang province and remains the oldest and most representative Chinese rice wine. During storage, Shaoxing Huangjiu is susceptible to environmental disturbance and produces colloidal haze to result in turbidity. In this study, the main composition and source of colloidal haze protein in Shaoxing Huangjiu were analyzed by two‐dimensional electrophoresis and matrix‐assisted laser ionization time‐of‐flight tandem mass spectrometry (MALDI‐TOF/TOF MS). The results showed that the proteins in colloidal haze mainly consisted of oat protein b1, oat‐like protein, di‐amylase inhibitor, pathogenesis‐related protein, pathogenesis‐related protein‐4, chitinase II derived from wheat and oat‐like protein, and beta‐amylase derived from rice. The amino acid composition and secondary structure of haze protein and supernatant protein in Huangjiu were further explored by high‐performance liquid chromatography and Fourier transform infrared spectroscopy. The study has broadened knowledge of the main composition and source of colloidal haze protein in Shaoxing Huangjiu. The corresponding results indicated that the amino acid composition from colloidal haze had the main characteristics of high hydrophobicity and low water solubility.


| INTRODUC TI ON
Huangjiu (Chinese fermented rice wine), a popular and traditional alcoholic beverage in China, is a kind of colloid solution with complex composition and rich nutrition. Generally, Huangjiu is produced from rice with "wheat Qu" and yeast by simultaneous saccharification and fermentation, following filtration, clarification, and sterilization. During Huangjiu brewing, the degradation of raw materials by co-fermentation with yeast, lactic acid bacteria, and fungi contributes to the production of abundant proteins, amino acids, vitamins, mineral elements, and oligosaccharides (Lv et al., 2016). Thus, Huangjiu is honored as "liquid cake" for its rich in nutrient substances (Yang et al., 2019). However, since Huangjiu could be affected by light, vibration, and oxygen during storage, its colloidal balance is broken to further result in the loss of light, flocculation, and colloidal haze (Van Sluyter et al., 2015).
In recent years, many researchers have become concerned about the colloidal haze components of Huangjiu (Lin, BAI, & ZOU, 2005;Xie, Meng, & Zhou, 2002). The study on the colloidal haze components in altar and bottled Huangjiu showed that the crude protein content of colloidal haze in the altar was 34.66% and that in the bottled Huangjiu accounted for 50.56% (Yang, Zeng, Chen, & Xiao, 2003). Similarly, our previous study revealed that the crude protein content of colloidal haze reached 50.60% in bottled Huangjiu, and further, the large molecular protein accounted for 72.62% of the total protein content (Xie, Meng, et al., 2002). According to the above studies, it was shown that protein was the main component of colloidal haze of Huangjiu. However, the increasing large proportion of large molecular weight proteins in colloidal haze did not decrease the stability of Huangjiu (Jiao, Xu, & Jin, 2017;YANG, YU, & WEI, 2005). This was speculated to relate to certain types of proteins in Huangjiu that were prone to colloidal haze.
In order to understand which types of proteins contribute most significantly to haze, two-dimensional electrophoresis (2-DE) and matrix-assisted laser desorption ionization time-of-flight tandem mass spectrometry (MALDI-TOF/TOF MS) were used to trace the types and sources of colloidal haze in Shaoxing Huangjiu. The amino acid composition and secondary structure of the colloidal haze in Shaoxing Huangjiu were further analyzed by high-performance liquid chromatography (HPLC) and Fourier transform infrared spectroscopy (FTIR), respectively. This research is to help understand the colloidal haze of Shaoxing Huangjiu and provide new theoretical guidance for solving the problem of colloidal haze.

| Preparation of haze protein from Huangjiu
The wine sample of 500 ml was stored at room temperature for a few days until cloudy precipitation occurred. Then, the sample was centrifuged (12,000 g, 30 min) to collect sediment, suspended in ultra-pure water, and the above process was repeated once. The clean sediment was dissolved in 2% aqueous ammonia, salted out with 80% saturated ammonium sulfate, then desalted in 1,000 Da dialysis bag at 4°C for 48 hr, and lyophilized to obtain haze protein samples (Pocock, Alexander, Hayasaka, Jones, & Waters, 2007).

| Haze protein identification by MALDI-TOF/ TOF MS
The protein band visible on the gel was cut out and added 200 μl of 30% acetonitrile solution containing 100 mmol/L ammonium bicarbonate. The mixture was shaken until colorless and dehydrated twice by adding 50 μl of anhydrous acetonitrile to obtain white micelles. Each tube was added 5 μl trypsin solution and left for 30-60 min at 4°C, so that the colloidal particles completely absorbed the enzyme solution. The excess enzyme solution was aspirated, and 20 μl of 25 mmol/L ammonium hydrogen carbonate solution was added to it, and the mixture was incubated at 37°C for 20 hr. The enzymatic hydrolysate was aspirated and transferred to a new centrifuge tube. The extraction solution was added to the original tube to ultrasonic extract. The enzymatic hydrolysate and the extract were rotary-concentrated in vacuo and then re-dissolved by adding 3 μl of TA60 solution (acetonitrile 600 μl, 10% trifluoroacetic acid (TFA) 10 μl, double-distilled water 390 μl). A sample of 0.7 μl was aspirated, and then, 0.7 μl substrate was added. After drying, the sample was subjected to peptide mass fingerprint analysis using a mass spectrometer. The positive ion reflection mode and the automatic data acquisition mode were set for the primary mass spectrometry data acquisition. The scanning range was 700-3500 Da.
Five to ten primary mass spectrometry peaks with good signal intensity were selected for secondary mass spectrometry analysis. Bio Tools software and Mascot software (http://www.matri xscie nce.com) were used, respectively, for integration and mass spectrometry data search in National Center for Biotechnology Information (NCBI).

Determination of free amino acids
The measurement method was determined according to Gao et al. (2011). Samples were filtered through micropore film (0.45 μm) before high-performance liquid chromatography (HPLC) analysis.
The relevant instrument and analysis parameters were as follows: TAG amino acid analysis column (3.9 mm i.d.×150 mm height), detect wavelength 254 nm, temperature 38°C, the solvent flow rate 1.0 ml ⁄min, and the injection volume 10 μl. Solvent A of mobile phase comprised 60 ml of acetonitrile and 940 ml of 0.14 mol/L sodium acetate (pH 6.40, containing 0.05% triethylamine). Solvent B of mobile phase was 40% water and 60% acetonitrile by volume. A gradient elution procedure with solvent A and solvent B was carried out: 0 min, 5% B; 15 min, 20% B; 35 min, 40% B; 42 min, 65% B; 50 min, 80% B; 52 min, 5% B; and 60 min, 5% B, and run time was 60 min.

Determination of total nitrogen content
Concentrated hydrochloric acid was added into 5 ml Huangjiu samples, then hydrolyzed at 110°C for 22 hr, transferred to volumetric bottles, filtered, steamed and centrifuged, and determined according to Kjeldahl method by means of K2300 Kjeldahl Nitrometer (Gao et al., 2011).

Hydrophobicity analysis of amino acids
The hydrophobicity of amino acids was calculated according to Lozano's empirical formula, and the average hydrophobicity of proteins was further calculated according to the following formula (Lozano, Combes, & IBORRA, 1994).
X i represents the molar ratio of an amino acid; Hφ i represents the hydrophobic value of an amino acid.

Sample pretreatment
After Huangjiu samples (500 ml) that have produced colloidal haze were laid aside for 5 days, the supernatant was separated siphoned, and then approximately, 80 ml of wine samples was left at the bottom. On the one hand, ammonium sulfate was added into the supernatant sample to 80% saturation. After overnight under 4°C, precipitate was collected by centrifugation, and the precipitate was transferred into the dialysis bag. After dialysis for 48 hr, the sample was freeze-dried and analyzed by FTIR.
On the other hand, samples left at the bottom were shaken well and then centrifuged at 12,000 g for 30 min to collect the precipitate. The precipitate was packed into 1,000 Da dialysis bag and dialyzed in flowing water for 48 hr. After freeze-drying, the precipitate was analyzed by FTIR.

FTIR analysis
Spectral processing was performed according to the method of Shevkani, Singh, Kaur, and Rana (2015). Infrared spectra were recorded using FTIR spectrometer. The spectrums were subjected to Fourier self-deconvolution (FSD), second derivative (SD) analysis, and curve-fitting procedures to locate peaks in amide I region using Peak Fit 4.12 software (Systat Software Company). The relative proportions of different secondary structures were determined by computing areas of spectral components (Gaussian peaks) assigned to a particular substructure in the amide I region.

| Two-dimensional electrophoresis analysis of haze protein in Huangjiu
The images of two-dimensional electrophoresis (Figure 1) showed that the distribution of proteins in haze protein from different Huangjiu samples was similar, mainly including acidic proteins and low molecular weight alkaline terminal proteins. The protein spots distributing between acidic and alkaline terminals were fewer and light in color, which indicated that the content of these proteins in haze protein was fewer. Low molecular alkaline terminal protein spots were darker in color, suggesting a higher content of such proteins.

| Identification of haze protein by MALDI-TOF/ TOF MS
The protein spots in Figure 1 were extracted for mass spectrometry analysis and retrieved in the NCBI database. A total of 48 spots were identified successfully and labeled in Figure 1. The information of each protein spot was shown in Table 1. A total of nine kinds of proteins were identified successfully in haze protein ( proteins derived from rice were oat-like protein and β-amylase with a molecular weight of 33.4 kDa and 55.4 kDa, respectively. Considering the above, the haze protein was mainly derived from wheat and rice. The dimer alpha-amylase inhibitor contributed the most significantly, and all were derived from wheat. The pathogenrelated protein (PR, PR-4, chitinase II) contributed the most significant proportion after alpha-amylase inhibitor, followed by oat-like protein and beta-amylase.
It was previously reported that dimer alpha-amylase inhibitor and oat protein were also the components of beer haze protein (Joshi, Panesar, Rana, & Kaur, 2017). Alpha-amylase inhibitor was first discovered in wheat seeds and was found to be sugar hydrolase inhibitor, which had a strong inhibitory effect on alpha-amylase from different sources (Silano et al., 1975;Svensson, Fukuda, Nielsen, & Bønsager, 2004 (Haque et al., 2014;Naz, Bano, Wilson, Guest, & Roberts, 2014;Wang et al., 2010). PRs are a class of monomeric proteins with a relative molecular weight of 10-40 kDa, and most of the PRs can tolerate low pH, heavy metal, protease, and high temperature (Gamir et al., 2017).
According to their amino acid sequence similarity, PRs can be divided into 14 families of PR-1-14. Among them, chitinase II belongs to the PR-3 family, which can degrade the cell wall of pathogenic bacteria and improve the disease resistance of plants. Chitinase was also found in wine turbidity (Iimure & Sato, 2013). Therefore, it was possible that because of the PR stability, PRs could not be removed even after being decocted and filtered during the brewing process.
As a result, PRs eventually were retained in the finished wine and were able to contribute to perceived turbidity in the Huangjiu.
Oat-like protein mainly found in wheat endosperm is a low molecular weight gluten. It is divided into subtypes A and B (Cao et al., 2018). Of the 168 amino acids in the subtype A protein, 36 were glutamic acid units, accounting for 21% of total amino acid. In the B-type protein, on the other hand, 80 of the 284 amino acids were glutamic acid, amounting to 28% of the total. This was consistent with a higher proportion of glutamic acid content in the following amino acid analysis of haze protein (Table 3). In addition, oat-like proteins containing highly redundant cysteine residues and disulfide bonds within and between molecules can form polymers, which have an important impact on the formation of turbidity (Hosseini, Kadivar, & Shahedi, 2012;Mirmoghtadaie, Kadivar, & Shahedi, 2009).

| Amino acid analysis of haze protein
The amino acid components of haze protein and supernatant protein in Huangjiu were analyzed, and the results were displayed in Table 3.
The hydrophobic value indicated the hydrophobicity of the amino acid, and the larger the value, the stronger the hydrophobicity. In the haze protein, glutamic acid content was the highest, accounting for 20.48% of the total amino acid content, followed by proline and aspartic acid, accounting for 10.12% and 8.54% of the total amino acid content, respectively. Among the supernatant protein, the highest content was also glutamic acid, accounting for 31.98% of the total amino acid content, followed by aspartic acid and serine, accounting for 8.67% and 7.53%, respectively. The hydrophobic value of hydrophilic glutamic acid and aspartic acid was equal to zero. And the hydrophobic value of serine was −0.3, which was the most hydrophilic amino acid. The content of serine in the supernatant protein was 7.53%, which was 1.

| Secondary structure analysis of haze protein
Regardless of the conformation of the side chains and the spatial arrangement of the entire peptide chain, the secondary structure of a protein refers to the spatial structure of the polypeptide chains.
Common secondary structures include α-helix, β-sheet, β-turn, and random curl (Zhao, Chen, Xue, & Lee, 2008). FTIR, an effective way to study the relationship between protein conformation and function, has been widely used in the study of protein secondary structure. As displayed in Figure 2, the absorbance of the haze protein was significantly higher than the supernatant protein in the band of 4000-3200/cm, while in the amide I band (1700-1600/cm), the absorbance of the supernatant protein was greater than the haze protein ( Figure 3). So, it was suggested that there were differences in protein structure between haze and supernatant protein.
The absorbance data of amide I band in the spectrogram were intercepted, and the spectrogram was fitted by Peak Fit 4.12 software (Systat Software Company). An obvious difference between the haze and supernatant protein was observed in the amide I band fit map ( Figure 4, Figure 5). Seven distinct peaks were obtained in haze protein, while there were only five peaks in the supernatant protein.
The secondary structure corresponding to each peak and the relative percentages of the secondary structures in the haze and supernatant protein were compared in Tables 4 and 5, respectively. Consistent with Figures 4 and 5, an obvious difference between the haze and supernatant protein in the secondary structure was revealed. For example, the supernatant protein did not contain α-helix structure, while the α-helix content in the haze protein was 15.91%. Moreover, the content of β-sheet, random curl, and β-turn in the supernatant protein was 26.9%, 25.3%, and 2.4% higher than that of the haze protein, respectively.
It has been reported that protein secondary structure was related to surface hydrophobicity (Van Dijk, Hoogeveen, & Abeln, 2015). In addition, the water solubility was negatively correlated with the proportion of α-helix content and positively correlated with the proportion of β-sheet and random curl, but no correlation with the proportion of β-turn content (Gao et al., 2011;Qu et al., 2018;Wang et al., 2011).
Considering the above results and discussion, the main structural features of haze protein in Huangjiu were the high content ratio of α-helix, the low content ratio of β-sheet and random curl.
In the production of Huangjiu, the sterilization was usually carried out under the conditions of 80-85°C for about 40 min in order to ensure the biological stability of bottled Huangjiu (Yang et al., 2019). It was possible that high-temperature sterilization destroyed the high-grade structure (including the secondary structure) of the protein, resulting in enhanced hydrophobicity of the protein, reduced water solubility, and eventually turbidity.

| CON CLUS IONS
The main composition and source of colloidal haze protein in Shaoxing Huangjiu were analyzed. Six kinds of proteins derived from F I G U R E 3 Amide I band Fourier transform infrared spectrogram for haze protein and supernatant protein F I G U R E 4 Curve-fitted map of secondary structure of haze protein in amide I band F I G U R E 5 Curve-fitted map of secondary structure of supernatant protein in amide I band wheat and two kinds of rice proteins were identified using MALDI-TOF/TOF MS in the haze protein, including oat-like protein b1, oatlike protein, dimeric alpha-amylase inhibitor, pathogenesis-related protein, pathogenesis-related protein-4, chitinase II, and beta-amylases. The molecular weight of the haze protein in Huangjiu mainly ranged between 13-16 kDa and 28-55.4 kDa. Amino acid analysis of haze protein revealed that the content of glutamic acid was the highest in haze protein, followed by proline and aspartic acid. Moreover, the average hydrophobic value of haze protein amino acids was 0.92, which was 16.46% higher than that of supernatant protein amino acids in Huangjiu. Besides, FTIR analysis showed that the content of α-helix structure in haze protein was 15.91%, while the supernatant protein did not contain α-helix. However, the contents of β-fold and random curl structure in haze protein were significantly lower than that in supernatant protein. It indicated that the high hydrophobicity and low water solubility of amino acid compositions were the main characteristics of the proteins which formed the colloidal haze in Huangjiu.

CO N FLI C T O F I NTE R E S T
The authors declare that they do not have any conflict of interest.

E TH I C A L R E V I E W
This study does not involve any human or animal testing.

I N FO R M E D CO N S E NT
Written informed consent was obtained from all study participants.